Print this page
Hybrid process design for analytics

Hybrid process design for analytics

Written by 

A renewed approach to embed analytics into complex business core processes

Written by Constantin Schröder and Tristan Poetzsch

Complex core business processes (like order to cash or purchase to pay) in most companies as well as their supporting IT-infrastructure evolved over time and grew complex as well as volatile. That is the reason why these processes are very hard to fit into the analytics world. Good analytics need to be designed into the system it should provide insights about and is therefore invasive. But redesigning core processes is costly and risky. Therefore, approaches like process mining try to generate information about processes in a passive format without invading the system. This, however, has the following limitations:

  1. Important data may not be available,
  2. data may require substantial preprocessing and cleaning,
  3. data may be suboptimally available (e.g. spread across different databases with different update cycles or writing schemes) and
  4. wrong conclusion might occur by just adhering to the process logs since complementary knowledge working might actively skip steps.

This stages the main issue for which we seek a solution. As business models and processes are currently evolving (“digital transformation”) and a lot of core processes are embedded in improved or even new IT solutions, it seems necessary to design these solutions in a way that they also optimally integrate analytics. On the other hand, this requires business engineers or process managers who design the systems to take a new, hybrid approach between functional business design and design-to-analytics, which we will refer to as Hybrid Process Design.

Disclaimer: We call the person who conceptualizes the functional requirements for the software a business engineer based on how the role was called by customers. In a more agile development, the role of a business engineer may be a part of the developer.

What do I as a business engineer need to do differently in Hybrid Process Design compared to standard process design?
Firstly, taking into account the functional business needs is still necessary. This can be for instance the logistics customer needs to input the number of tons as well as start and end point. But secondly, it is important to talk to the business analyst and analytics department to understand their analytics needs (e.g. it is necessary to know what number of cargo tons is passing a specific border). This requires additional workshop concepts with business and data analysts within new frameworks. Such phenomenon needs to be translated to architectural and functional designs of the software.

How can I structure a workshop with architects, business and data analysts?
It may be helpful to structure your workshop along some guiding questions:

  1. How does the specific process fit into the overarching business process?
  2. What questions come up frequently from management boards and how can these be answered?
  3. What departments are engaged within the process at which point?
  4. What are the overall relevant metrics of the whole business process relevant to the management (e.g. throughput time, runtime, non-value adding steps)?
  5. How can the relevant metrics be made available best?
  6. How can the currently designed process part provide better data (in terms of update cycles, error handling, data format)?
  7. How is the data currently stored? How is data from other software parts and systems along the whole business process stored and formatted?

What skills does this require from me?
There are basically three new contexts that a hybrid business engineer needs to master:

  1. A broader understanding of the business from a management perspective to aid in the definition of relevant insights about the processes for the management and business analysts.
  2. A basic knowledge of data science and especially the cleaning of data to design data inputs and functions in such a way, that it is optimally usable for analytics. This does mean for example that most user inputs should be validated, or associated data formats should be stored similarly formatted.
  3. A better understanding of data storage is necessary. From optimizing RDBMS not only for minimal redundancy, but also for optimal analytics usage to using data lakes for more flexible storage, a business engineer needs to translate and balance functional, architectural and analytic needs for data storage. The following skill matrix may help:

 

Tristan Poetzsch

Computer aided cognition and AI specialist, currently working at Nexgen Business Consultants.

Media